IP Library › Granted Patent US 12681170

Granted Patent B1

US 12681170 · App. 17/828,995 · Granted Jul 14, 2026

Radar-based ground plane determination

Inventors: Patrick Blaes (Redwood City, CA); Badeea Ferdaous Alferdaous Alazem (Redwood City, CA)

Assignee: Zoox, Inc.

G01S13/931 G01S13/89 G01S2013/93271

View Patent ↗

Loading inventors, assignments & file history…

Monitor This Case

Get email alerts when status or documents change.

Order Certified Copies

Most orders are placed with the USPTO same day — all within 24 business hours.

Order via The Patent Place →

Pre-filled with this patent's details

Quick Facts

Patent No.: US 12681170
App. No.: 17/828,995
Granted: Jul 14, 2026
Kind: B1

External Links

USPTO Patent Center↗ Full Patent PDF↗

Abstract

Techniques for estimating a ground plane based on lidar data and/or attributes of the ground plane are discussed herein. A vehicle captures radar data, e.g., 4D radar data including height information, as it traverses an environment. The radar data can include direct returns from an object and reflected or multipath returns, e.g., that reflect off a ground surface and the object. A position of the ground plane can be estimated based at least in part on a distance between direct returns and the reflected returns. Attributes of the ground plane may be determined from differences between the direct returns and the reflected returns.

Claims (71)

1 . A vehicle comprising:

a radar sensor disposed on the vehicle;

one or more processors; and

one or more non-transitory media storing instructions that, when executed by the one or more processors, perform operations comprising:

receiving, from the radar sensor, radar data containing a plurality of radar returns corresponding to an environment of the vehicle, individuals of the plurality of radar returns having associated range information, height information, and velocity information;

determining a first subset of the plurality of radar returns comprising one or more direct returns from an object in the environment;

determining, based at least in part on the height information, a second subset of the plurality of radar returns spaced vertically below the first subset of the plurality of radar returns;

determining a correspondence of an attribute of the first subset of the plurality of radar returns to an attribute of the second subset of the plurality of radar returns;

determining, based at least in part on the second subset being spaced vertically below the first subset and the correspondence of the attribute of the first subset of the plurality of radar returns to the attribute of the second subset of the plurality of radar returns, that the second subset comprises reflected returns reflected off the object and a road surface, the road surface associated with a first height information in the environment;

inputting, into a model, the second subset comprising the reflected returns independent of requiring the first subset comprising the one or more direct returns to be input into the model;

receiving, as an output from the model, an estimation of a ground plane in the environment, the estimation including a second height information of the ground plane; and

determining, based at least in part on the estimation, a variation between the second height information of the ground plane and the first height information of the road surface.

2 . The vehicle of claim 1 , wherein the determining the estimation of the ground plane comprises:

generating a first histogram associated with the first subset of the plurality of radar returns, the first histogram being based at least in part on the range information or the velocity information;

generating a second histogram associated with the second subset of the plurality of radar returns, the second histogram being based at least in part on the range information or the velocity information;

determining the correspondence of the range information or the velocity information; and

determining the estimation of the ground plane based at least in part on the first histogram and the second histogram.

3 . The vehicle of claim 2 , wherein the determining the ground plane comprises determining a midpoint along a vertical dimension between a first position associated with the first histogram and a second position associated with the second histogram.

4 . The vehicle of claim 1 , wherein the operations further comprise:

determining, based at least in part on a vertical distance between the first subset of returns and the second subset of returns, a pitch of the ground plane.

5 . The vehicle of claim 1 , wherein the operations further comprise: generating, based at least in part on the estimation of the second height information of the ground plane in the environment, one or more controls for controlling the vehicle; and controlling the vehicle based at least in part on the one or more controls.

6 . A method comprising:

receiving, from a radar sensor, radar data comprising a plurality of radar returns corresponding to an environment of a vehicle, the plurality of radar returns comprising one or more direct returns from an object in the environment and one or more reflected returns reflected off the object and a road surface, and individual of the plurality of radar returns having associated range information, height information, and velocity information;

determining a first subset of the plurality of radar returns based at least in part on an attribute of the first subset of the plurality of radar returns;

determining a second subset of the plurality of radar returns based at least in part on an attribute of the second subset of the plurality of returns;

determining a correspondence of the attribute of the first subset of the plurality of radar returns to the attribute of the second subset of the plurality of radar returns;

determining, based at least in part on the height information, that the first subset of the plurality of radar returns vertically aligns with the second subset of the plurality of radar returns;

determining, based at least in part on the correspondence of the attribute of the first subset to the attribute of the second subset and on the first subset vertically aligning with the second subset, that the first subset comprises the one or more direct returns and that the second subset comprises the one or more reflected returns;

inputting, the first subset comprising the one or more direct returns and the second subset comprising the one or more reflected returns into a model;

estimating, by the model, a position of a ground plane in the environment at a vertical location between the first subset and the second subset; and

determining a difference in height between the ground plane and the road surface.

7 . The method of claim 6 , wherein:

the attribute of the first subset of the plurality of radar returns comprises at least one of a range or a velocity; and

the attribute of the second subset of the plurality of radar returns comprises at least one of a range or a velocity.

8 . The method of claim 6 , further comprising:

generating a plurality of representations of the radar data, individual of the plurality of representations being associated with an individual segment of a plurality of discrete segments of the environment; and

determining the first subset of the plurality of radar returns and the second subset of the plurality of radar returns based at least in part on the plurality of representations of the radar data.

9 . The method of claim 8 , wherein the plurality of representations of the radar data comprise a plurality of histograms based at least in part on the range information or the velocity information.

10 . The method of claim 8 , further comprising:

comparing vertically-aligned representations of the plurality of representations;

determining, based on the comparing, the first subset of the plurality of the radar returns as returns corresponding to a first representation of the plurality of representations and the second subset of the plurality of the radar returns as returns corresponding to a second representation of the plurality of representations, the second representation being vertically below the first representation; and

determining the position of the ground plane as a midpoint between the first representation and the second representation.

11 . The method of claim 8 , wherein the plurality of discrete segments correspond to sections of a two-dimensional grid partitioning the environment.

12 . The method of claim 6 , wherein the determining the position of the ground plane comprises:

inputting the radar data into a machine learned model; and

receiving, from the machine learned model, an estimation of the ground plane.

13 . The method of claim 12 , further comprising:

training the machine learned model using known or annotated ground plane information.

14 . The method of claim 13 , wherein the known ground plane information is based at least in part on a map of the environment.

15 . The method of claim 6 , wherein the position of the ground plane is a first position of the ground plane associated with the object, the method further comprising:

determining one or more additional positions of the ground plane based at least in part on radar data associated with one or more additional objects; and

generating, based on the first position of the ground plane and the one or more additional positions of the ground plane, a representation of the ground plane in the environment.

16 . The method of claim 15 , wherein the representation of the ground plane is a three-dimensional mesh of the ground plane.

17 . One or more non-transitory computer-readable media storing instructions that, when executed, cause one or more processors to perform operations comprising:

receiving, from a radar sensor, radar data comprising a plurality of radar returns corresponding to an environment of a vehicle, individual of the plurality of radar returns having associated range information, height information, and velocity information, and the plurality of radar returns comprising one or more direct returns from an object in the environment and one or more reflected returns reflected off the object and a road surface;